What you are actually tracking
AI Overviews are the written summaries Google sometimes places at the top of its results, with links to the pages it drew from. They do not appear for every search. For the ones where they do appear, three questions matter to you, and each needs its own record.
Does an overview appear at all for this search? If it does not, there is nothing to be cited in. For many searches it will not. For some it comes and goes.
If it appears, does it cite your site? A link to one of your pages in the sources shown is the clearest sign you were used.
Who else is cited, and what does the summary say? The other sources are your real competition for that search, and they are often not the businesses you compete with commercially. Directories, review sites and publications show up a lot.
Tracking is the measuring half of the work. What you change in response is a separate topic, covered in our guide to how to rank in AI Overviews. And if you want the measuring, writing and publishing handled together, our overview of a combined AI search engagement shows how that works.
One reminder before the methods. AI Overviews change. The text may be different an hour later, and the sources can be reordered. So every method below is a sample, and the goal is a pattern across many questions and many weeks, not a verdict from one search.
Tracking by hand: a routine that costs nothing
You can learn a great deal with a spreadsheet and an hour a month. Doing it by hand also means you read every summary, which no dashboard makes you do.
Build the question list
Write 20 questions a buyer would type before choosing a business like yours. Mix the kinds: broad ones like what is the best way to do X, specific ones with a place or a product in them, comparison ones and problem ones like why does my X keep doing Y. Resist the urge to use only your own product names, because buyers who do not know you do not type them. Save the list and do not change it.
Search cleanly
Use a private window, signed out of your Google account, with your location set to where your customers are. A signed-in search is shaped by your own history, and you search for your own business so often that Google may treat you as someone interested in it. Using the same browser setup each month keeps the comparison fair.
Log what you see
For each question, record the date, whether an overview appeared, whether your site was linked in it, the other sources linked, and a line on how the summary described the topic. Take a screenshot too. A screenshot is evidence when a result changes and you need to remember what it used to say.
Repeat on a schedule
Once a month is enough for most businesses. Run the same list the same way. If an overview did not show on your first pass, check again later, because Google does not show them for every search every time.
Read across the log
After three months, look for patterns. Which kinds of question trigger an overview? Which of your pages get cited, and what do they have in common? Which sources keep showing up where you do not? The third question is the most valuable.
What Search Console can and cannot tell you
Google Search Console is the free report Google gives site owners on how their pages perform in search. It is the only place to get real numbers from Google about your site, so it is worth knowing how it treats AI Overviews.
As of this writing, Google counts appearances and clicks from AI Overviews inside the normal Web search totals in the Performance report. There is no separate filter that isolates them. Google's documentation can change, so check the current help pages before relying on this.
What that means in practice:
You can see the questions you appear for. The queries report lists the searches that showed your pages. Question-shaped searches are the ones most likely to trigger an overview, so they are worth reading closely.
You cannot tell which impressions were inside an overview. An impression on a search where an overview appeared looks the same as any other.
A fall in clicks on a question page is a clue, not proof. If impressions hold steady and clicks slide on a question you used to win, an overview may be answering some people before they click. Check the search by hand to see whether an overview now appears.
Use Search Console to pick which questions to put on your manual list. Find the question-shaped searches with the most impressions, add them, and check them by hand each month. That joins the real demand figures Google gives you to the direct look at the result that no report provides. Our glossary entry on Google Search Console has a short explanation if you have not set it up.
Tracking with tools
Software helps once the list is long or the work repeats across locations. Tools that track AI Overviews generally work in one of two ways. Some are rank trackers that added a field showing whether an overview appeared on a search and which sources it listed. Others are AI visibility platforms that cover overviews alongside other assistants.
Both are sampling tools, so judge them on how they sample. Our guide to answer engine optimization tools covers that in detail. For overviews in particular, ask these:
- Where and how is the search run? Results differ by location and device. A tool that cannot set a place may not match what your customers see.
- How often does it re-check? Overviews appear and disappear, so a single check can miss one.
- Does it show the cited sources and the summary text, or only a yes or no?
- Does it keep history? The trend is the asset.
- Can you export the data?
Check any tool against your hand log for a month. If they tell the same story, you can trust the tool for the next hundred questions. If they do not, ask the vendor why before paying.
What tracking misses, and how not to be fooled
A few traps catch people who start tracking.
Reading one result as the truth. The same search can produce a different summary tomorrow. Judge by the pattern over weeks.
Counting being cited as being chosen. A link in an overview's sources is a good sign, but sources are often listed in a row, and the summary may lean on a different one. Read the text.
Tracking only your own name. Searches that include your business name tell you what Google thinks of you once it knows you. The searches that matter most for growth are the ones where a buyer does not yet know you exist.
Forgetting the other assistants. AI Overviews are one place buyers ask. Many ask a chat assistant directly. A business absent from overviews may still be well placed there, and the reverse. Our overview of AI visibility monitoring covers the wider set.
Expecting a sales number. A person who sees you in an overview may click, or may search your name later, or call. Almost none of that is labeled. Treat tracking as a measure of presence, and judge the business result in your normal sales figures over a longer period.
Finally, do not rush to fix every gap. If you are cited for 6 of 20 questions in month one, and 8 of 20 in month three, that is movement. Document it, keep going, and look for what the successful pages share before changing everything.
What your log should look like
A spreadsheet is enough. Use one row per question per check, with these columns.
- Date of the check.
- Question, copied exactly from your fixed list.
- Location and device used.
- Overview shown? Yes, no, or loaded after a delay.
- You cited? Yes or no, and which page.
- Other sources cited, by site type, such as directory, publication, competitor or forum.
- Summary note, one line on what the summary said.
- Screenshot file name.
An example row, invented to show the format: the date, the question how long does a kitchen remodel take, location set to a mid-sized city, overview shown yes, cited no, other sources cited two publications and a directory, note says cabinet lead time sets the schedule, screenshot saved.
After a few checks, add a summary tab with the share of questions where an overview appeared and the share where you were cited. Keep those as two separate numbers. The first tells you how much of your territory Google is answering with a summary. The second tells you how you are doing within it.
The column that pays off most is the one for other sources by type. After a few months, a pattern usually shows. If publications and directories fill most of the slots, those are the places to be listed or mentioned.
Local businesses and online stores track differently
The method is the same, but two kinds of business need small adjustments.
Local businesses
Local searches are shaped by where the person is. Google often shows a map and local listings, and an overview may or may not appear above them. Set your location to the area you serve, and if you cover several areas, track each as its own set of rows. A result in one town tells you nothing about the next.
Track both things on a local search: whether an overview appears and whether you are in the map listings. They draw on different signals, such as your business profile and reviews, so a business can do well in one and be absent from the other.
Online stores
Product searches often show shopping results, and overviews tend to appear more on informational and comparison searches than on a plain product name. Split your list into questions about choosing a product, which are where a summary is likely, and searches for a specific item, which may not show one.
For stores, an overview that cites a buying guide or a comparison page you wrote is a good sign that your informational pages are doing their job. If you are weighing how to build those pages, our overview of the AI visibility checklist lists the steps.
What to do when tracking says you are not cited
Tracking is only useful if it leads to a next step. When your log shows an overview that cites other sources and not you, work through these questions in order.
Do you have a page that answers this question? If not, that is the gap, and writing one comes first.
If you do, is it indexed? Check in Search Console. A page Google has not indexed cannot be cited.
Does the page state the answer early and plainly? Compare its opening to the passage the overview used. Often the cited page simply gets to the point faster.
What kinds of sources are cited instead? If they are directories and review sites, being listed there is part of the answer. If they are other businesses, read what they say that you do not.
Fix one thing, note the date and check again in a month. The steps to improve a page are in our guide on how to rank in AI Overviews. Do not try to change everything at once, because then you cannot tell what worked.
How RunOctopus handles this
A tracking log shows where Google skips you. RunOctopus writes and publishes the pages that give it a reason to cite you.
RunOctopus is a done-for-you service. Its AI, Ollie, researches your business, then writes and publishes pages directly on your own website. A team plans, writes, reviews and publishes alongside Ollie. It works with Shopify, WordPress, WooCommerce and Wix.
There is nothing to install, nothing to approve, nothing to upload and nothing to schedule. You do not approve each page. Pages go live on their own. In the first week you get 8 long-form articles, 6 pages and a custom interactive tool built for your niche. After that it is 175+ pieces of content a month, written, published and linked. You also get reports on where ChatGPT, Perplexity, Claude and Gemini name your business.
There is a 90-day promise: more visibility on Google and in AI answers within 90 days, measured against the 90 days before you joined, or every dollar back. Paste your address into the free site read and Ollie reads the site and shows the plan. No account, no call.